PRomotion Of Physical activity through structured Education with differing Levels of ongoing Support for people at high risk of type 2 diabetes (PROPELS): study protocol for a randomized controlled trial

Detailed advice leaflet (Control Group)

Participants in the control group receive an advice leaflet detailing the likely causes,
consequences, symptoms and timeline associated with prediabetes, as well as information
about how physical activity can reduce the risk of developing T2DM. This leaflet is
informed by Leventhal’s Common Sense Model 41], which also underpins the structured education programme. Participants also continue
to receive standard care from their GP.

Structured education programme followed by annual group maintenance sessions (Walking
Away Group)

Participants are given the same advice leaflet as the control group and invited to
attend an updated version of the PREPARE structured education programme within three
months of their baseline clinic visit, titled Walking Away from Type 2 Diabetes. Participants
are also offered annual maintenance education sessions, revisiting the key messages
of the initial Walking Away programme, and discussing any benefits and barriers they
have experienced to increasing their physical activity.

The Walking Away programme fully incorporates the curriculum and content of the successful
PREPARE programme 32], 37], but was renamed and updated to incorporate current terminology and guidance 6] and now includes an accredited educator training and quality assurance pathway. The
PREPARE programme has previously been shown in a single centre RCT to increase ambulatory
activity by 2000 steps per day compared to control conditions over a 12-month period
when combined with pedometer use with sustained improvements in glucose levels. 32], 37]

The programme is delivered by two trained educators to groups of up 10 participants
(who are welcome to bring a guest). The content of the programme, examples of activities
and the underlying theoretical structures are presented in Table 2. The key aim of the programme is to increase participants’ physical activity predominantly
through increased walking, and to prompt the use of self-monitoring and feedback using
pedometers. During the programme participants use their daily habitual step count
(measured at baseline prior to the education programme) to set personalised activity
goals. They are provided with a pedometer (Yamax SW200) as a self-monitoring tool.
Typically, sedentary participants are encouraged to increase their activity levels
by at least 3000 steps per day, equivalent to around 30 minutes of walking. Those
achieving more than 6000 steps per day are encouraged to try to reach at least 9000
steps per day, an amount that is likely to include 30 minutes of walking activity
in addition to usual daily activity 32]. Those achieving more than 9000 steps per day are encouraged to at least maintain
their current activity levels and develop further goals should they wish to. Goal
attainment is encouraged through the use of proximal objectives, such as increasing
ambulatory activity by 500 steps per day every two weeks. Participants are encouraged
to make an action plan detailing where, when and how their first proximal goal will
be reached, to repeat action planning for each new proximal goal, to wear their pedometer
on a daily basis and to self-monitor their ambulatory activity using a specifically
designed steps-per-day diary. A full description of the rationale, development and
efficacy of the work underpinning Walking Away are detailed elsewhere 32], 42].

Table 2. Outline of the Walking Away programme (delivered to the Walking Away group and the
Walking Away Plus group)

Following the initial group session, participants are offered annual group-based maintenance
sessions at 12, 24 and 36 months, each lasting 2.5 hours. These maintenance sessions
are designed to re-visit the key messages of the first session, and build self- efficacy
through the sharing of successes, problem solving barriers, goal setting and self-monitoring
using pedometers. This mode of delivering behaviour change maintenance is designed
to be directly relevant to primary care pathways where annual clinical follow-up of
those with a high risk of chronic disease, such as prediabetes, is recommended 5], 6].

Structured education programme plus follow on support (Walking Away Plus)

Participants receive the same advice leaflet, Walking Away structured group education
programme and annual maintenance sessions. In addition they are introduced and given
access to a highly tailored text-messaging follow-on service designed to support behaviour
change and pedometer use with additional telephone calls from educators that are initiated
after attendance at the Walking Away programme. Details of the text messaging and
telephone follow-on support service are detailed in Table 3 and have been submitted for publication elsewhere. In brief, approximately one week
after the Walking Away session, an educator telephones the participant to help them
identify a short- and long-term step goal and action plan for the next six months,
and to elicit information on tailoring variables including confidence in increasing
physical activity, previous experience of physical activity and potential mobility
issues that prevent walking being the primary activity. The educator records the information
in an online form and saves it on a database for use by the text-messaging programme.
A key feature of the text-messaging and pedometer support is that participants are
prompted to text in their weekly step count. They then receive automated feedback
by text message tailored to the extent to which they achieved their goals, in relation
to the tailoring variables. Participants also receive text-messages using evidence-based
behaviour change techniques and a six month telephone call to review their goals and
action plans, prompt problem solving and provide social support. The key techniques
used in the follow-on support are goal setting (behaviour), action planning, self-monitoring,
goal review, problem-solving and social support 43]. The intensity of the ‘follow-on support’ varies over the course of the year (see
Table 3) and is repeated over the four years of the study. Participants who do not wish to
receive text-messages can opt out at any time.

Table 3. ‘Follow on’ support for the Walking Away Plus Group, repeated over the four years
of intervention

Educator recruitment, training and quality assurance (intervention fidelity)

Educators can be registered health care professionals (e.g. nurse, dietician) or a
suitable non-registered professional (e.g. Health Trainer). Across the study sites,
16 educators have been recruited from a range of registered and non-registered Health
Care Professionals from local healthcare providers and other appropriate settings.
This mix of personnel was included in order to make the study as pragmatic and generalisable
as possible to routine health care systems. To ensure that all interventions are delivered
by educators as planned, a fidelity standard operating procedure has been put in place.
All educators attended an initial two day training course to ensure that they understood
the theories and philosophy that underpin the Walking Away programme and the content
and resources used within it. All educators were given a written curriculum to support
their delivery of the programme and given the opportunity to practise delivery of
the programme. Quality assurance of delivery is undertaken to assess educational style
and content using established tools used within routine care through the DESMOND Collaborative
44]. Educators receive structured and instructive feedback from their assessor and key
goals and action plans are developed in order to help them improve their performance.
Prior to the delivery of the one year intervention, educators received training to
be able to deliver the Walking Away Maintenance session which is also supported by
a written curriculum.

Educators attended additional training in delivering the telephone calls to participants
in the ‘Walking Away Plus’ group. The training was supported by an extensive curriculum
outlining the contents (behaviour change techniques, patient-centred communication
skills) and mode of delivery of the follow-on support. The curriculum also contains
standardised scripts that educators are asked to follow, an explanation of the intervention
fidelity procedures, and standardised reflection sheets and checklists to promote
and assess fidelity of the phone calls. Educators are asked to audio-record the telephone
calls, listen back to a sample, complete the checklists, and discuss these with the
intervention lead. The number of calls assessed and their frequency depends on the
competence level of the individual educators as well as the year of intervention delivery.
Monitoring the delivery of the text-messaging system is done on a weekly basis by
examining automatically generated lists of messages sent and received.

Study outcomes

Data collection clinics are run by research nurses in the Leicester Diabetes Centre,
the MRC Epidemiology Unit, Cambridge and other local community centres and clinic
areas. All staff have been trained in study procedures and data are collected following
standardised operating procedures. Written informed consent is obtained from all participants
prior to the commencement of data collection. Details of all clinical assessments
and outcome measures are provided in Table 4. Participants are sent a letter with details of selected clinical results after each
visit, and their results are also sent to their general practitioner.

Table 4. Participant assessments at each time point

Primary outcome

The primary outcome measure is change in ambulatory activity (steps per day) at 48 months,
assessed by accelerometer (Actigraph GT3X+) with an intermediary assessment at 12 months.
Participants are asked to wear the accelerometer on a waistband (in the right anterior
axillary line) for seven consecutive days during waking hours following their baseline
and follow-up visits. In addition, participants are asked to fill in a log sheet each
day that they wear the accelerometer indicating time the accelerometer was taken off
at night, time they went to sleep, time they woke up and time the accelerometer was
attached. At the end the seven days of wear, participants are asked to return the
accelerometer and the log sheet to the research team in a pre-paid envelope. Raw acceleration
data are captured and stored at 100 Hz. Data processing will be undertaken on commercially
available analysis tools. At least three days valid wear is required to count as a
valid recording. Non-wear time is determined by one hour or more of consecutive zero
counts. Due to the potential for bias between groups in factors used to acquire valid
accelerometer data, average wear time and the number of valid days will be included
as covariates in the analysis.

Secondary outcomes

Accelerometer

The accelerometer used to measure the primary outcome, detailed above, is also used
to assess the number of censored steps taken per day, defined as steps taken above
an intensity used to distinguish between purposeful and incidental ambulation 45]. In addition, commonly used cut-points will be used to distinguish between time spent
sedentary and in time spent in light-, moderate, and vigorous-intensity physical activity
46]. We will also use this data to assess compliance to the physical activity recommendation
of undertaking at least 150 minutes of moderate-to-vigorous intensity physical activity
per week in bouts of at least 10 minutes.

Time spent lying, standing, sitting and postural transitions

Participants are also asked to wear an activPAL3â„¢ device for the same seven days as
the ActiGraph accelerometer. This is a small slim thigh worn monitor that uses accelerometer-derived
information about thigh position to determine body posture (i.e., sitting/lying, standing
and stepping). The activPAL3™ is initialised using manufacturer’s software with the
default settings (i.e., 20Hz, 10s minimum sitting-upright period) and participants
are asked to wear the device continuously (24 hours/day). The device is covered with
a nitrile sleeve and fully wrapped in one piece of waterproof dressing (Hypafix Transparent)
to allow participants to wear the device during bathing activities. The activPAL3â„¢
is worn on the midline anterior aspect of the upper thigh and secured using hypoallergenic
waterproof dressing (Hypafix Transparent).

Recent Physical Activity Questionnaire (RPAQ)

Self-reported physical activity is measured using the RPAQ. This assesses physical
activity across four domains (domestic, recreational, work, commuting) over the previous
month. The domestic section contains questions regarding computer use, TV-viewing
and stair climbing at home. The questions in the recreational domain ask about frequently
performed activities, including frequency and duration. The work domain examines the
level of physical activity associated with their current employment, and commuting
assesses four modes of usual transport: walking, cycling, car, and public transport.
It has shown moderate-to-high reliability for physical activity energy expenditure,
and good validity for ranking individuals according to their time spent in vigorous
intensity physical activity and overall physical activity energy expenditure 47]. Participants also complete the Neighbourhood Environment Walkability Survey (NEWS)
which captures the environmental context in which participants live and will be used
to assess environmental determinants of physical activity and physical activity behaviour
change 48], 49].

Biochemical variables

Standard biomedical outcomes are assessed by venous sampling. These consist of HbA1c,
lipid profile (triglycerides, HDL, LDL, total cholesterol), urea and electrolytes
(sodium, potassium, urea, creatinine) and liver function tests (albumin, total bilirubin,
alkaline phosphatase, alanine transaminase).

At the Leicester site only, participants are assessed for fasting and 2-hour post
challenge glucose and insulin levels using an Oral Glucose Tolerance Test (OGTT).
The OGTT results will be used to provide greater clinical insight into how the promotion
of physical activity affects metabolic health in prediabetes, given 2-hour glucose
and insulin levels better reflect peripheral insulin sensitivity. Given that the new
WHO guidelines 40] have led to local clinical practice basing diagnosis of T2DM on HbA
_1c
criteria, the routine use of the OGTT has been phased out of primary care. Therefore,
in order to comply with local guidelines and to avoid confusing clinical management
strategies in recruited practices, OGTT samples taken in Leicester are frozen and
will be analysed after the final study visit (48 month) and not form part of a diagnosis
of diabetes. The OGTT involves a fasting blood sample being taken from the patient
before they are then given a glucose load of 75 g. A second sample is taken 2 hours
later. Fasting and 2-hour plasma glucose and insulin samples are taken at each clinical
visit and stored in a ?80 °C freezer using standardised, stable methodology within
the Leicester Diabetes Research Centre.

Genetics

A blood sample for genetic analysis is also collected in those who provide their consent.
The aim of this sample will be to investigate group level associations and interactions
of physical activity, obesity and genes in the development of T2DM. The genetic assessments
will be focused on genes for which there are biological plausibility for interaction.

Anthropometric and demographic variables

Body weight, body fat percentage, height and waist circumference are measured to the
nearest 0.1 kg, 0.5 %, 0.5 cm and 0.1 cm respectively. Waist circumference is measured
using a soft tape mid-way between the lowest rib and iliac crest. Arterial blood pressure
is obtained from the right arm of the seated participant. Three measurements are taken
and an average of the last two measurements will be used. Total upper arm length,
forearm length, total leg length and lower leg length are measured on the left side
of the body. Information on ethnicity, medication history, current smoking status,
family history of diabetes in first and second degree relatives, and muscular/skeletal
injury that prevents physical activity are obtained by self-report.

Cardiovascular risk

Cardiovascular risk is calculated using the Framingham risk calculator.

Self-reported dietary behaviour

In order to capture dietary behaviour, two short questionnaires used in previous research
studies by our group are administered to the participants for self-completion. These
questions are based on dietary questionnaires developed for the EPIC (European Prospective
Investigation of Cancer and Nutrition) study and the international NAVIGATOR (Nateglinide
And Valsartan in Impaired Glucose Tolerance Outcomes Research) study 50], 51].

Sleep

Participants self-report on two questions concerning sleep duration over the past
24 hours and over a usual week. There is accumulating evidence for an association
between short sleep duration (6 hours per 24 hours) and long sleep duration (?10 hours
per 24 hours) and metabolic dysfunction 52].

Health related quality of life

Health related quality of life data is measured using the European Quality of Life-5
Dimensions (EQ-5D) 53] and the Short Form (SF-8) Health Survey 54]. The EQ-5D is a standardized questionnaire that was developed for use as a measure
of health outcomes and defines health in terms of five dimensions: mobility, self-care,
usual activities, pain or discomfort, and anxiety or depression. It is widely used
to calculate ‘quality adjusted life years’ (QALYs) which are essential to cost-effectiveness
analysis. The SF-8 is a self-administered questionnaire measuring eight health domains
(general health, physical functioning, role physical, bodily pain, vitality, social
functioning, mental health and emotional roles) with eight questions. The standard
(4 week) recall format will be used. Data from SF-8 is represented as a physical component
score and a mental component score.

Depression and anxiety

Depression and anxiety are measured using the Hospital Anxiety and Depression (HADS)
Scale, to produce independent subscales for anxiety and depression 55].

Health resources

A health resources questionnaire records the number of times over the past 12 months
that the participant has seen a health care practitioner such as a GP, nurse or other
health workers, and the number of times they have been to hospital. In addition, the
contact and costs associated with the intervention will be captured by the research
team. These costs will be inputted into the cost effectiveness analysis of the intervention.

Process measures

Perceptions of diabetes risk

Perceptions and perceived knowledge of diabetes risk is measured at 12 and 48 months
using the validated Brief Illness Perceptions Questionnaire (BIPQ) 56]. This eight item instrument uses an 11 point Likert scale (0?=?no effect, 10?=?complete
effect) to measure five cognitive diabetes risk representations (consequences, timeline,
personal control, treatment control, and identity), two emotional representations
(concern and emotion) and risk comprehensibility (perceived knowledge).

Self-efficacy in relation to physical activity

Self-efficacy is measured at baseline, 12 and 48 months. Six items measure participants’
confidence in their ability to undertake any form of moderate- to vigorous-intensity
physical activity for 10 minute periods, increasing incrementally from 10 minutes
to one hour each day. Items use a 100 % confidence rating scale (where 0 %?=?no confidence,
and 100 % represents complete confidence). An overall score is calculated by summing
the efficacy scores for each time period and dividing by the number of time periods.

Enactment of the behaviour change techniques

Participants’ use of behaviour change techniques included in the Walking Away and
Walking Away Plus groups is assessed at 12 and 48 months. A 5-point Likert scale (where
1?=?most of the time and 5?=?never) assesses how often participants set goals, form
action plans, use a pedometer, complete an physical activity log, are aware of their
activity levels, and are trying to be more physically active.

Uptake and adherence to Walking Away and Walking Away Plus interventions

Measures of uptake and adherence to the intervention groups will include: 1) group
session attendance to Walking Away, 2) group session attendance to the annual maintenance
sessions at 12, 24 and 36 months 3) proportion of phone calls completed, 4) number
of participants who registered for the text messaging service, 5) number of STOP messages
received for test messaging (i.e. number opting out of the text messaging and pedometer
support), 6) proportion of intended texts sent, and 7) number of step count texts
received from participants relative to the number of requests they are sent (engagement).

Qualitative process evaluation

A process evaluation will be conducted using a combination of ethnographic methods
including observations of education sessions and individual interviews. Briefly, we
will observe a sample of Walking Away sessions and undertake interviews with a sample
of participants from all three trial groups. Analysis will be informed by the constant
comparative method; the focus will be exploring whether and how participants engage
with each component of the intervention(s).

Sample size

Primary outcome

For 1-beta?=?0.8, alpha?=?0.025 (allowing for 2 a priori comparisons against control
conditions), SD?=?4000 steps/day and a drop-out of 30 %, we require 436 per group
(1308 in total) to detect a 1000 steps/day difference in change in ambulatory activity
(equivalent to 10 mins walking/day or 70 mins walking/week) between the intervention
groups and control group. Assuming 25 % of participants in the total cohort are SA
we have an 80 % power to detect around a 2000 steps/day difference when comparing
two intervention comparisons to the control group (alpha?=?0.025) in the SA population.

Several intervention studies with a follow-up of between 3 to 12-months reported a
standard deviation of change in ambulatory activity of around 3000 to 4000 steps per
day in individuals with T2DM, prediabetes or in sedentary individuals 32], 57]–59].Therefore we have anticipated a standard deviation of change of 4000 steps per day.
An intervention effect of 1000 steps was considered the minimum clinically significant
difference between groups and equates to around a 4 % difference in the risk of cardiovascular
morbidity and mortality 13].

Secondary outcome

Given that around 95 % of the general population fail to meet the Chief Medical Officer’s
physical activity guidelines when measured objectively by accelerometers 60], this study also allows for the 10 % difference in those meeting the current physical
activity recommendation to be detected at follow-up based on 1-beta?=?0.8, alpha?=?0.025.
Consistent with the calculation for ambulatory activity, this study has sufficient
power (1-beta?=?0.8, alpha?=?0.025) to measure a 10 minute/day difference in change
in the time spent in moderate-to-vigorous intensity physical activity based on previous
work undertaken by our group 16].

This study has sufficient (1-beta?=?0.8, alpha?=?0.025) power to allow for clinically
meaningful differences for change in the biochemical measures to be detected in the
entire study cohort and after stratification by ethnicity; fasting glucose (0.3 mmol/l),
2-h glucose (1 mmol/l) and HbA
_1c
(0.25 %).

Furthermore assuming a conversion rate to T2DM in the control group of at least 24 %
over the course of the entire study (4 years), we will have an 80 % power to detect
a 40 % reduction in the relative risk of T2DM in both intervention groups compared
to the control group. The estimated conversion rate is at the lower level reported
for traditionally defined prediabetes 7], 61]. We anticipate that the inclusion of an HbA
_1c
defined prediabetes in this study will act to marginally lower the conversion rates,
whilst the inclusion of a large South Asian group will act to increase the conversion
rates.

Data analysis

Analysis will involve two a priori comparisons; both intervention groups will be compared to the control group. Should
any of these comparisons reveal a significant difference, then a third a priori comparison will be undertaken by comparing the difference between intervention groups
– this will be included as a secondary analysis. Multiple linear regression analysis
will be used to investigate the differences in the change in physical activity level
achieved between groups at 48 months, after adjusting for potential areas of bias
between groups, valid accelerometer wear time and the number of valid wear days. Analysis
will be conducted on the cohort as a whole and stratified by ethnic group; interaction
terms will be used to quantify the effect of ethnic group. We will also use interaction
terms to assess whether the effects of the interventions are modified of gender, age,
ethnicity, family history of T2DM or whether prediabetes status was confirmed at baseline;
significant interactions will followed by stratified analysis. We will also assess
change in physical activity at 12 months as a secondary outcome. The primary analysis
will be based on analysing those with complete data at each time point in the group
to which they were randomised. A sensitivity analysis will be undertaken to assess
the impact of imputing missing data through multiple imputation and through a per-protocol
analysis by removing those that failed to attend the initial Walking Away session
in both intervention groups; per-protocol analysis for the Walking Away Plus group
will be defined as removing those that that failed to attend the initial Walking Away
session OR those that failed to register for or actively stopped their text messaging
support via the STOP function within the first 2 months.

Analysis of secondary biochemical and anthropometric outcomes will be analysed using
the same strategy and at the same time-points as that described for the primary outcome.
Differentials in the time to T2DM between groups will be plotted using the Kaplan–Meier
method for comparing survival curves; the log rank test will be used to assess for
differences between the groups.

Health economics

We will undertake a costing exercise to determine the cost of delivering the initial
interventions covering expenditure such as educator time, educator training and quality
assurance. In addition we will determine the cost of the follow-up maintenance support
group-sessions and the staff and other costs of the individually tailored telephone
and text messaging package for maintenance support. Resource use incurred will be
costed using actual costs in the trial and/or standard references for unit costs such
as Unit Costs of Health and Social Care 62]. In addition to the primary endpoint, we will analyse the within-trial impact of
the interventions during the trial on other outcomes that are pertinent to the long-term
economic analysis, i.e. use of antihypertensives and lipid-lowering therapies, blood
pressure, health utility and incidence of T2DM.

Long-term costs and benefits of the interventions will be evaluated through a combination
of the within-trial outcomes and decision-analytic modelling to simulate long-term
incidence ofT2DM, microvascular complications arising from T2DM and cardiovascular
events. Specifically for progression to T2DM, estimating long-term progression will
require a statistical model built partly from incidence data from the trial. The underlying
incidence curve will be based on rates of progression in the control arm, and a survival
model with time-varying hazards will be built to demonstrate the effect of a unit
change in physical activity on risk of T2DM over time. This will allow the impact
of alternative assumptions about the degree of maintenance of physical activity beyond
the 4-year follow-up period to be modelled. The underlying progression of T2DM beyond
the 4-year follow-up will be estimated by the above 4-year survival curve and assumptions
about medium-term maintenance of physical activity, but also informed by the trajectory
of survival curves from long-term diabetes prevention studies such as the Finnish
Diabetes Prevention study 63].

An important input for the modelling will be the effect of increasing physical activity
on cardiovascular risk. The relationship between changes in physical activity and
cardiovascular risk will be incorporated into an existing decision-analytic model
of prevention of diabetes. This relationship will be informed by a recent study that
specifically calculated the effect of change in ambulatory activity (steps per day)
on cardiovascular disease risk 13]. The model will be chosen from existing models used previously for work for NICE
64] on prevention of diabetes or developed as part of the School for Public Health diabetes
prevention theme within ScHARR, funded by the National Institute for Health Research
(NIHR). The economic model aggregates the costs of the intervention, prescribed medications,
therapy, the costs of one-off treatments (e.g. cost of amputation), and on-going treatment
of complications (e.g. treatment following stroke). The cardiovascular risks of participants
with prediabetes, or with diabetes will be estimated using the UKPDS risk engines
65], 66]. A further adjustment will be made so that the risk can be adapted for the South
Asian population using evidence advised by clinical colleagues.

Separate evaluations will be undertaken for the overall group and for the South Asian
subgroup. Cost-effectiveness will be reported in terms of incremental costs and QALYS
and the incremental cost-effectiveness ratio. Uncertainty around the results will
be explored through probabilistic sensitivity analysis and related techniques for
identifying the most important drivers of uncertainty, which will be presented on
a cost-effectiveness acceptability curve and a cost-effectiveness plane.

Trial Steering Committee (TSC) and Data Monitoring and Ethics Committee (DMEC)

The trial is overseen by TSC comprising an independent chair, independent clinical
and academic members and the principal investigator. This committee is responsible
for the overall management and oversight of the trial. The steering committee are
blinded to all information regarding treatment assignments until after the database
is locked for final analysis.

A fully independent DMEC reports to the TSC. This comprises an independent chair and
a statistician. The DMEC are responsible for the interests of participant safety and
data integrity. The DMEC will undertake safety data reviews every nine months, unless
otherwise deemed necessary. In addition the DMEC will also review analysis plans.